Metrological contributions to magnetic resonance-based electric properties tomography

Magnetic resonance-based electric properties tomography (EPT) is a quantitative imaging technique that, starting from acquisitions performed with a magnetic resonance scanner, aims at producing a map of the distribution of the electric properties at radiofrequency within a human subject. All the EPT implementations proposed in the literature deal with the problem of noise amplification. This makes the noise in the retrieved maps a major issue for EPT development. Indeed, in order to make EPT results really quantitative in the strict sense of measurement science, it is fundamental to find a way to evaluate the actual uncertainty with which the electric property values are retrieved in each pixel. In this paper, a method based on Helmholtz-EPT that estimates at the same time the conductivity and its uncertainty in each pixel is proposed. The method is applied to acquisitions of a heterogeneous phantom and possible applications of the uncertainty map are shown.